Table 6.12 includes a summary of the results obtained and/or analysed in this research. As confirmed in previous studies (Santero et al. 2011a; Santero and Horvath 2009), the traffic delay and the PVI Rolling Resitance components can have a relevant impact in the life cycle
142 of a pavement, compared to other phases. If, in this study, this is always true for the PVI Rolling resistance phase, for the work zone impact, the methods and the methodological assumptions make the results span a big range.
Traffic delay
In both case studies, the simplified approach with the HCM provides larger values of CO2, than the microsimulation model. This difference is marginal in the case study with a low
volume of traffic (A17) and gets significantly bigger for the A1(M), characterized by a higher volume of traffic.
The results are sensitive to all the analysed parameters: traffic volume, TM strategy, EF and road network boundary extension.
The increase in traffic volume generates a rise in the CO2 emissions. This increase,
based on the assumptions made, spans between 66.7% and 410.3% for the A17 case study and 33.2% and 3201.9% for the A1(M) case study, depending on the model used to calculate the impact of the work zone.
If in the A17 the impact of the increase in traffic volume is bigger with the Aimsun microsimulation model, the opposite is true for the A1(M), where the impact of this variable, for an increase of 30% over the original value, generates results over 30 times bigger. In addition, in this last case, once the traffic impact increases (from 10% to 30 %), the increase of CO2 emissions is very large.
The TM strategy selected can strongly impact the results, both in terms of timing and layout. For the A17, the Base case scenario represents the TM option with lowest emissions. Scenario 2, involving an earlier closure of the road, produces an increase of the CO2 emissions
between about 6-7 times bigger than the base case scenario (the increment is comparable for the two models). Scenario 3 with the road closure and extensive traffic diversion has the highest emissions of CO2. In the A1(M) case study, Scenario 2, requiring the use of the hard
shoulders as a running lane and providing larger road capacity, has the lowest impact, although it lasts longer than Scenario 3 (detour).
The EF selected to estimate the CO2 vehicle emissions significantly affect the results;
by using the MOVES EF the CO2 emissions are about three times that using the EFT EF for
both case studies.
The definition of the area of impact of the work zone is relevant in the traffic modelling process. The small network, compared to the Mini network, estimates larger emissions. By
143 contrast, in the Big network, the extra emissions estimated are smaller than in the Mini network.
PVI Rolling resistance
In both case studies, the use of these models provides considerably different results for all the components. It amounts to one order of magnitude for the basic and the total component, while the deterioration component is positive in the UCPRC model and negative in the VTI model. This means that in the VTI model, the total component is smaller than the basic component, due to the reduction of the deterioration component over the years. The opposite is true for the UCPRC model.
Regardless of the deterioration in the IRI and MPD over the analysis period, the two models return considerably different results (10272 ton against 1170 ton in the A17 and 109344 ton against 18058 ton in the A1(M).
In the sensitivity tests, the results are affected by all parameters, though to different degrees. While the traffic growth and the emission factor parameters affect the results, the combined impact is not significant overall. However, in this study the impact of the traffic volume and fleet on the pavement deterioration (IRI and MPD) was not taken into account, due to the lack of available models. This may affect the actual influence of the traffic growth on the vehicle emissions.
The CO2 emissions due to the pavement roughness are very sensitive to the pavement
surface deterioration over time. In fact, for both models, not only is the range of potential impact due to the PVI wide, the lowest emissions in the two models occur under different pavement deterioration rate scenarios (no deterioration in the UCPRC model and average deterioration in the VTI model).
144 Table 6.12: Summary of the results
Original Construction Simapro Maintenance 2009 Simapro Aimsun HCM Aimsun HCM Aimsun HCM Aimsun HCM Aimsun HCM Scenario 3 (Detour) Aimsun EF Moves HCM *** B D T B D T UCPRC 1170 217 1387 18058 4586 22645 VTI 10272 -600 9,672 109344 -4205 105,139 UCPRC 1,288 225 1,513 19,696 4,767 24,462 VTI 10,515 -657 9,858 110,887 - 4,575 106,311 UCPRC 1,340 261 1,601 20,702 5,480 26,182 VTI 11,748 -709 11,039 125,056 - 5,032 120,023 UCPRC 1,162 203 1,365 17,773 4,290 22,064 VTI 10,395 -651 9,744 109,517 - 4,530 104,987 UCPRC 1,127 186 1,313 17,225 4,012 21,237 VTI 9,241 -562 8,679 97,503 - 3,841 93,662 UCPRC 1,413 247 1,660 21,597 5,242 26,840 VTI 11,358 -703 10,654 119,971 - 4,894 115,077 UCPRC 1,288 - 1,288 19,696 - 19,696 VTI 10,515 - 10,515 110,887 - 110,887 UCPRC 1,288 1,242 2,530 19,696 21,214 40,910 VTI 10,515 558 11,073 110,887 5,216 116,102 * from (Spray 2014)
** from (Galatioto et al. 2015) *** B: Basic; D: Deteriorration; T=Total
47.31 837.00 Year/Event 96.58 5,194 143.08 10.45 29.09 52.76 4,031 223 48.58 329.84 67.74 1,913 10% Sensitivity test Pavement Deterioration No Worst
Use phase Rolling resistance 2009-2029
Base case scenario
Average case scenario
No Sensitivity test Emission Factors 10% Sensitivity test Traffic growth No 63.29 2.40 Sensitivity testTraffic Volume 10 **3.43 4.00 20 **4.67 4.90 30 **9.99 Construction Raw materials/ transport/ onsite equipment
Base case scenario *702
Base case scenario *370
Traffic delay Maintenance 2009
Base case scenario
**1.94 6.30 7.95 Sensitivity testTraffic Management Scenario 2 16.46 17.93 TOTAL RESULTS
Activities Variables Model
A17
CO2(e)(ton)
A1(M)
CO2(e)(ton)
145
7 Discussion and Conclusion
In this chapter the implications and impact of the results obtained in this study and described in Chapter 1 are discussed in the context of the literature review in Chapters 2, 3 and 4 and the methodology described in Chapter 5. The main aim of this chapter is to describe how the study has met the research aims and objectives stated in section 1.2.1 and how it has answered some outstanding questions and filled some significant research gaps, as highlighted in section 4.6.
While the first section describes the overall conclusions of the study, the other two sections discusses in details specific ‘key’ considerations of the work zone traffic delay impact and rolling resistance components.
7.1 Pavement LCA
As discussed in section 2.3, the implementation of LCA principles in the pavement domain is complex, due to some methodological issues that reduce the accuracy and reliability of the results obtained. For long time, the traffic delay and the rolling resistance components, whose impact can be significant under specific conditions, have been omitted from previous pavement LCA studies. In the last years, the research has made progress, either assessing the impact of these components and developing or using models able to explain their behaviour. The results obtained in this research have, first of all, confirmed the relevance of these two components in the life cycle of a road pavement compared to other phases. Therefore, as already stated in previous studies, performing pavement LCA assessment, withouth taking into account the work zone traffic delay and the PVI rolling resistance components, may lead to incorrect, or at least incomplete, conclusions.
An important outcome from this thesis is the significant influence of the models applied to analyse both the traffic delay and the rolling resistance. To assess the impact of the traffic delay during construction and maintenance events, existing traffic and emission models have been used with different level of sophistication, which has a significant impact on the LCA results. Therefore, the choice of the traffic and emission models needs to be based on the study objectives and on the available resources. The estimation of the impact of the
146 Rolling resistance on the vehicle emissions requires the development of models to estimate the deterioration of the pavement surface properties (both in terms of IRI and MPD) over time and to correlate them with the and with the vehicle fuel consumptions. Currently, there are few models available in the literature, which are calibrated for site-specific conditions. The results are sensitive both to the model used to estimate the PVI rolling resistance CO2
emissions, and to the surface deterioration rate chosen. Site specific elements and methodological choice affect the development of the rolling resistance and fuel consumption models, meaning they are not suitable for all geographical locations.
The selected model is not the only source of uncertainty in the assessment of these components, requiring specific methodological assumptions. These, as shown in this study, can have a relevant impact on the results, generating a high level of uncertainty. The traffic delay results are sensitive to all the input variables considered in this study: the traffic growth, the TM strategy adopted, the EF model and the extent of the road network modelled around the work zone.
For the rolling resistance, if the deterioration rate is a significantly sensitive parameter, the traffic growth and the EF/fuel efficiency predictions, combined to predict future vehicle emissions, have a relatively small effect because they cancel out to a large extent. Changes in predicted future traffic levels or EF could change this result and should be kept under review. These research outcomes highlight the importance of incorporating uncertainty into pavement LCA. The reliability and accuracy of an LCA is affected by the reliability of the assumptions, methodologies and models adopted. LCA results should not be presented as ’single figure’ absolute values, but rather as a range of values to estimate the uncertainties and variability that lie behind them.
7.2 Work zone traffic delay
Chapter 3 has shown that the analysis of traffic delay during maintenance activities in pavement LCAs is generally based on macroscopic analytical models or microsimulation models, whose main features have been described in the same chapter.
The level of sophistication of the model used to assess the impact of the work zone during maintenance events has a significant impact on the LCA results, as described in section 6.2.1. For both case studies and approaches used, the HCM provides higher values of CO2
147 and deceleration of the vehicles in the work zone. This significant dissimilarity of the results may be due to specific features of the HCM models that determine an overestimation of the additional fuel consumption caused by the traffic delay.
For the A1(M), the macroscopic approach with the HCM involves queuing during much of the day (average speed of 8 mph), while the microsimulation model does not. As explained in section 5.4.1.3, the calculation of the emissions with the HCM_LCCA model is strongly affected by the speed of the queuing traffic, since a small change in this value in congested conditions (between 0 and 25 mph) can generate significantly different results, in terms of fuel consumption and CO2 emissions. However, the curve used in the LCCA
procedure to calculate this value does not allow a precise and accurate evaluation of the queue speed, reducing the reliability of the outcome of the model. In order to use this model in a confident way, this procedure should be updated by using more accurate methods of evaluating queue speed and associated emissions.
Instead, in the A17 case study, the bigger value of the emissions with the HCM approach may be due to an underestimation of the real value of the saturation flow, which affect the number of vehicles passing at green during the traffic light cycle and, therefore, the length of the queue and the idling time.
Other factors explaining the results obtained are the different emissions factors used in the two models, the different approaches used to estimate the queue speed (in Aimsun, it is based on other factors, such as the car-following models) and the fact that in the HCM, several parameters (such as Capacity and average speed) are based on empirical data on USA roads. When traffic models are used, therefore, these model input parameters need to be accurately monitored and controlled, in order to provide a better understanding of the outcome of the study.
As stated in section 3.3, other potentially significant variables to take into account in the analysis of the traffic delay impact are the traffic volume considered for the study, TM strategy adopted, the EF model to convert the vehicle power in the fuel consumption and CO2
emissions and last, but not least, the extent of the road network modelled around the work zone. The results of this study have shown that, in different ways and with different level of impact, the final results are sensitive to all the chosen parameters. In particular:
- A greater value of AADT results in an increase in CO2 emissions due to the work
zone. However, based on the results obtained, it is not possible to draw a general conclusion related to the impact of the increased traffic volume on the results based
148 on the original AADT (low-medium-high level of traffic) or the model used. In fact, while in the A17 the impact of the increase of traffic volume is bigger with Aimsun microsimulation model, the opposite is true for the A1 (M). Instead, it is possible to state that, during congestion, the increase of tailpipe emissions is not directly proportional to the increase in the number of vehicles (AADT), since the congestion and increase of the traffic approaching the work – zone, will generate not only a rise in the number of vehicles that produce CO2 emissions at reduced speed, but also an
increase of the length of the queue.
- The sensitivity analysis has also shown that the TM of the road works, in terms of type, duration and timing, significantly affect the results, but not always in the same ways. For instance, road closure and diversion onto adjacent roads, which is usually the safest and most economical option (no interaction between opposite traffic flow or road workers), is not always an alternative with a lower carbon footprint.
- The EF model selected to convert the output of the traffic model into CO2 emissions
can vary the output results substantially. This is due, as seen in the section 3.2, to the different approach used to calculate the FC and emissions (average or instantaneous model) and to the fact that they have been developed in different countries and validated under different conditions.
- The extent of the road network modelled is a relevant factor in the analysis of the traffic delay component. During a maintenance event, the behaviour of the vehicles is influenced by the traffic in the work zone and they could select alternative route directions to reach the same destination point. In the worst scenario, the congestion generated in the work zone could extend beyond the modelling area. An incorrect evaluation of the area of impact may lead to an underestimation or overestimation of the actual value of the CO2 emissions.